Parameterized Cell Layout Optimization for VLSI Yield
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current automated solutions lack the capability to optimize yield in very large scale integration (VLSI) circuit design by modifying the physical shapes of parameterized cells, particularly for via cells with non-schematic parameters that affect manufacturing sensitivity.
Innovation Solution
A method that reads a physical design containing parameterized cells, creates new versions with optimized parameter settings, and updates parameter values to improve yield, using a mapping of parameter values to edge positions and applying mathematical optimization to adjust physical design shapes based on desired objectives and constraints.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If parameterized cells are used to reduce data volume and personalization complexity, then device complexity is reduced, but the capability to optimize yield through physical shape modification is lost
Solution Approach 1:
The patent segments the parameterized cell system into two distinct layers: schematic parameters (for logical functionality) and physical parameters (for manufacturing optimization). This segmentation allows independent optimization of each layer, enabling yield improvement through physical shape modification without increasing schematic complexity or data volume.
Solution Approach 2:
The patent introduces a new dimension of physical parameters beyond the traditional schematic parameters. By adding this spatial/physical dimension to the parameterized cell framework, the system enables optimization of manufacturing yield through physical shape adjustment while maintaining the original schematic-level abstraction and data efficiency.
2Ease of operation
If traditional parameterized cells are used with one-to-one correspondence between schematic parameters and layout parameters, then schematic control is simplified, but optimization of physical manufacturing sensitivity is limited
Solution Approach 1:
The patent divides the parameter system into schematic parameters (controlling logical behavior) and physical parameters (controlling manufacturing characteristics). This segmentation breaks the rigid one-to-one correspondence, allowing schematic simplicity to be maintained while independently optimizing physical manufacturing sensitivity through dedicated physical parameter controls.
Solution Approach 2:
The patent makes the parameter correspondence dynamic rather than static. The mapping between schematic and physical parameters can be adjusted based on optimization objectives, allowing the system to adapt the relationship between schematic control and physical optimization depending on the specific design goals and manufacturing requirements.
3Manufacturing precision
If via cells have many non-schematic parameters for physical manufacturing sensitivity, then manufacturing precision can be improved, but device complexity increases
Solution Approach 1:
The patent segments parameters into schematic and physical categories, allowing non-schematic physical parameters to be managed separately. This segmentation reduces parameter management complexity by organizing the many parameters into distinct groups with different optimization goals, making it easier to handle the increased number of parameters for manufacturing sensitivity.
Solution Approach 2:
The patent organizes the many non-schematic parameters into a structured physical parameter space with clear categories and relationships. By imposing this organizational dimension on the parameter set, the system manages complexity through systematic structure rather than through reduction in parameter count, enabling comprehensive manufacturing optimization.
Data Source
AI summary
A method of layout optimization containing parameterized cells includes reading a physical design containing parameterized cells, creating a new version for each of usage of a given parameterized cell. The method optimizes physical design shapes of each new version of the parameterized cell by assigning variables to parameters of the parameterized cell according to a desired objective. Then, the method updates the parameters of each new version of the parameterized cell and replaces each new version of the parameterized cell with an instance of the parameterized cell having updated parameters. The method can optionally adjust physical design shapes based on constraints related to the parameters.